The spectrum of V4334 Sgr (Sakurai's object) in August, 1998

6 Pages·0.45 MB·English

by  Ya. V. Pavlenko

A&A manuscript no. ASTRONOMY
(will be inserted by hand later) AND
Your thesaurus codes are: ASTROPHYSICS
08(08.01.1; 08.16.4; 08.09.2: Sakurai’s object)
The spectrum of V4334 Sgr (Sakurai’s object) in August,
⋆
1998
Ya.V. Pavlenko1,2 and H. W. Duerbeck3
1 Main Astronomical Observatory of Academy of Sciences of Ukraine, Golosiiv woods, Kyiv-127, 03680 Ukraine, e-mail:
[email protected]
1 2 Isaac Newton Instituteof Chile, Kiev Branch
0
3 WE/OBSS, Free UniversityBrussels (VUB),Pleinlaan 2, B-1050 Brussels, Belgium, e-mail: [email protected]
0
2
Received ; accepted
n
a
J
Abstract. Theoretical spectral energy distributions are 1995 1996 1997 1998 1999
4
1.0001000
1 computedforagridofhydrogen-deficientandcarbon-rich
V
model atmospheres with a range in Teff of 4000−5500 K
1 and logg of 1.0−0.0,using the technique of opacity sam- 12
v
pling, and taking into account continuous, atomic line
9
2 and molecular band absorption. The energy distributions 14
2 are compared with the spectrum of V4334 Sgr (Saku-
1 rai’s object) of August 12, 1998 in the wavelength inter-
16
0 val 300 − 1000 nm. This comparison yields an effective
1 temperature of V4334 Sgr of Teff = 5250±200 K, and a
0 18
value for the interstellar plus circumstellar reddening of
/
h EB−V =1.3±0.1.
p
20
-
o
r
t Key words: Stars: individual: V4334 Sgr (Sakurai’s ob- 22
s 0.00000
a ject)–Stars:AGBandpost-AGBevolution–Stars:model 49500 50000 50500 51000 51500
: atmospheres – Stars: energy distributions – Stars: effec- J.D. − 2400000
v
tive temperatures – Stars: gravities – Stars: interstellar
i
X reddening
Fig.1. The visual light curve of V4334 Sgr, according to
r
a Duerbeck et al. (2000). The times of five spectroscopic
observations,analyzedbyAsplundetal.(1997,1999)and
1. Introduction
Kipper & Klochkova (1997), are indicated by open trian-
V4334 Sgr, the “novalike object in Sagittarius”, was dis- gles. The times of our two spectroscopic observations of
covered by Y. Sakurai on February 20, 1996 (Nakano et April, 1997 and August, 1998 are marked by filled trian-
al. 1996). Soon after discovery, it was found to be a final gles. The second observation took place at the beginning
He flashobject,a staronthe rareevolutionarytrackthat of a deep brightness decline.
leads from the central stars of planetary nebulae back to
the red giant region, an object also called a “born-again
giant”.Its progenitorwasa faintblue star(∼ 21m)inthe century, V4334 Sgr is a very quickly evolving object
(Fig. 1). It shows striking similarities with the unfortu-
centreofalowsurfacebrightnessplanetarynebula(Duer-
nately poorly documented final He flash object V605 Aql
beck & Benetti 1996). Early spectroscopic studies of the
that erupted in 1918 – 1923 (see Clayton & de Marco
object found an increasing hydrogen deficit in the atmo-
(1997) for an overview). Between March 1996 and April
sphere,andaC/Oratio>1 (Asplundet al.1997,Kipper
1997, V4334 Sgr evolved from early F to late red giant
& Klochkova 1997).
stages.InfraredobservationsbyKimeswengeretal.(1997)
Unlike the slowly evolving born-again giant FG Sge,
and by Kamath & Ashok (1999)show a noticeable excess
which rose in brightness during the first half of the 20th
in the infrared K band starting March 1997, indicative
Send offprint requests to: Ya.V. Pavlenko of dust formation. However, only in early 1998 a ∼ 1m
⋆ Based on observationscollected attheEuropean Southern depression appeared in the optical light curve (Liller et
Observatory,La Silla, Chile al. 1998, Duerbeck et al. 1999, 2000). This moderate de-
2 Ya.V. Pavlenkoand H.W.Duerbeck:The spectrum of V4334 Sgr in August, 1998
pression persisted for a few months, and was followed by Table 1. Abundances of H, He, C, N, O (scaled to PN
i
a ∼ 6m decline which started in August 1998. After a = 1) used in this paper.
partialrecoveryin late 1998– early 1999,the dust obscu-
ration continued with increasing strength, and it may be Element Asplund Asplund Kipper& [C]=+0.6
assumed that in the year2000,only the short-wavelength et al. et al. Klochkova
tail of the cooldust envelope can be observedin the opti- (1997) (1999) (1997)
cal spectrum, while the stellar photosphere is engulfed in H −1.730 −2.42 −2.10 −2.45
the opaque dust shell. Unfortunately, not much is know He −0.027 −0.020 −0.01 −0.05
about these late phases, since the spectroscopic coverage C −1.73 −1.62 −2.01 −1.05
N −2.53 −2.52 −2.70 −2.55
is poor due to the extreme faintness of the object in the
O −1.93 −2.02 −2.59 −2.05
visible region.
A spectrum of V4334 Sgr was obtained on April 29,
1997,beforeanydustobscurationintheopticalregionbe-
came evident, but after the strong IR excess had evolved. Klochkova(1997).Theirresultsdifferinsomedetails,last
Thisspectrumwascomparedwiththeoreticalspectralen- not least because there are differences in the physical in-
ergydistributions(Pavlenko,Yakovina&Duerbeck2000). put parameters and in the details of the procedure.
The authors derived a temperature of Teff =5500 K, and The abundances of some elements obtained by As-
aninterstellarreddeningofEB−V =0.7.Inthispaper,we plund et al. (1997) for May 1996 are given in the sec-
apply the same methods to a spectrum which was taken ond column of Table 1. Asplund et al. (1999) obtained
15 months later. another set of abundances for October 1996. For com-
parison, we show the abundances of Kipper & Klochkova
(1997), obtained independently for a spectrum of July
2. Observations
1996. In general, their results differ more from Asplund
On August 12, 1998, a spectrum (range 355−1000 nm, et al.(1997)than the results ofAsplund etal. (1997)and
resolving power ∼ 2000) was taken by A. Piemonte with (1999). Therefore, they provide a more interesting base-
the B&C Cassegrain spectrograph attached to the ESO line for numerical experiments carried out to clarify the
1.52m telescope, andkindly put at the authors’disposal. dependence of our results on input abundances.
It was taken around the time of the onset of the ∼ 6m From the common point of view, hydrogen is an im-
decline, at a time when V4334 Sgr was obviously already portant element, and its abundance in the atmosphere
suffering ∼1m of visual dust obscuration (Fig. 1). of V4334 Sgr is steadily declining (Asplund at al. 1997).
The spectrum was reduced using standard IRAF rou- However, Pavlenko et al. (2000) showed that the depen-
tines, and the spectral energy distribution (SED) was de- denceofourresultsonlogN(H)ismuchweakerthanthat
rived using spectrophotometric standard stars. The ob- found by Asplund et al. (1997) for the early stages of its
served SED is influenced by interstellar reddening as well evoluton. Indeed, in the case of lower Teff (<6500 K) the
−
asbyadditionalcircumstellarreddening.The valueofthe contribution of H and H to the total opacity in the line
interstellarreddeningisstillnotaccuratelyknown;wewill forming region is reduced in comparison with “hot mod-
assumeEB−V =0.7inthefollowing,aswasderivedinthe els”,becausemolecularabsorptionwhichdoesnotinclude
analysis of the “dustless” spectrum of April 29, 1997, by hydrogen compounds dominates there.
Pavlenko, Yakovina & Duerbeck (2000); note, however, Furthermore,and more importantly, the input carbon
that Duerbeck et al. (2000) prefer EB−V = 0.8, a value abundance ofAsplund et al.(1997)is largerby up to 0.6
whichisalsoconsistentwiththeobservedversustheoreti- dex than that which was obtained by them from the fine
calBalmerdecrementoftheplanetarynebulasurrounding analysisofhighresolutionspectra.Thatisthewellknown
V4334 Sgr (Kerber et al. 2000). “carbonabundanceproblem”(Asplundetal.2000),which
We will assume that the dusty envelope acts like an has, unfortunately, not yet been resolved.
additional amount of interstellar reddening. Dereddened One may think that, in the meantime, the model at-
spectralenergydistributions ofthe August12,1998spec- mospheres should not be computed with the derived C
trum were calculated for assumed values 0.7 ≤ EB−V ≤ abundances by Asplund et al. (1999) but rather with the
1.3, and compared with model atmospheres. higher input values.The use of the derivedlowerC abun-
dance would result in a new “carbon problem” and thus
evenlowerC abundanceswouldbe derived.Toclarify the
3. Procedure and Results
case, we studied the impact of higher carbon abundances
Our computations of model atmospheres and synthetic on the model atmospheres and SEDs of V4334 Sgr. We
spectra are carried out in a classical approach: a plane- used 4 abundance sets (i.e. Asplund et al. (1997) for May
parallel model atmosphere in LTE, with no energy diver- 1996,KipperandKlochkova(1997)forJuly1996,Asplund
gence. We assumed chemical compositions of V4334 Sgr etal.(1999)forOctober1996,andthe “carbonrichcase”
as determined by Asplund et al. (1997) and Kipper & with[C]=+0.6withrespecttotheAsplundetal.(1999)
Ya.V. Pavlenkoand H.W.Duerbeck:The spectrum of V4334 Sgr in August, 1998 3
abundance)inordertostudydifferentaspectsoftheprob-
lem:
– to model the impact of abundance changes on our re-
sults,
– to avoid possible uncertainties related with the “car-
bon problem”.
Opacity sampling model atmospheres and syntheti-
cal spectra of V4334 Sgr were computed by the SAM941
(Pavlenko 1999) and WITA6 programmes, respectively.
The model atmosphere computation procedure is de-
scribedin Pavlenkoet al.(2000)and Pavlenko(2000).To
clarify the topic, some details of the computations proce-
dure are given here:
– Convection was treated in the frame of the mixing-
length theory with l/H = 1.6. Overshooting was not
considered in the computations;
– the list of atomic lines was taken from VALD
(Piskunov et al. 1995);
– molecular opacities were computed in the frame of
JOLA approach.We take into account the absorption
in the 20 band system of diatomic molecules (see Ta-
ble 2 in Pavlenko et al. 2000);
– bound-free absorption of C i, N i and O i atoms were
computed using cross-sections of Hoffsaess (1979);
– SAM941 and WITA6 use the same opacity lists.
Model atmospheres, synthetical spectra and SEDs
were computed for a microturbulent velocity of 5 km/s,
which is a typical value for atmospheres of post-AGB
stars.The resultingtheoreticalspectrawere convolvedby
a gaussian with a half-width of 0.5 nm.
Basically, the atmospheric structure of chemically
peculiar stars should respond to abundance changes
Fig.2. The T = f(P ) structure of model atmospheres
g
(Pavlenko & Yakovina 1994, 2000). To study the impact
computed for Asplund et al. (1997), Kipper & Klochkova
of the set of input abundances on our results, we carried
(1997),Asplundetal.(1999),and[C]=+0.6abundances.
outa few numericalexperiments with modelatmospheres
A second diagram shows their T =f(τ) structure.
computed with abundances of Asplund et al. (1999) and
Kipper & Klochkova (1997).
The comparison of the temperature structure of a – the structure of the outermost layers of model atmo-
model atmosphere Teff/logg = 5200/0.0 computed for spheres computed for the Asplund et al. (1997) and
the four abundance grids is shown in Fig. 2. We prefer to Kipper & Klochkovaabundances is similar. Moreover,
use pressure as a depth parameter, because temperature the differences of the computed SEDs are rather weak
structures in the coordinates (τross,T) look very similar (Fig.3).Morepronounceddifferencesarefoundforthe
(seeFig.2binPavlenkoetal.2000).Ingeneral,thephoto- “carbon-rich”case;
spheresareshiftedoutwardsforhighercarbonabundances – in the blue region, the “flux peaks” which correspond
due to the opacity increase. Let us note a few items: to the minima of molecular absorption, are less pro-
nouncedforthemodelbasedonKipper&Klochkova’s
– The structure of model atmospheres of V4334 Sgr, (1997) abundances, but even more so in the observed
computed for abundances by Asplund et al. (1999), spectrum (Fig. 3);
Kipper & Klochkova (1997), and the “carbon-rich – Wehavetoadmitthatthereisanobviouslackofcom-
case” differ substantially, especially in the deep lay- putedopacityintheblueregion,andthemodelisonly
ers. Differences of model atmospheres computed for correct in a qualitative sense (Pavlenko & Yakovina
Asplund et al.’s (1999) abundances obtained for May 2000).
and October of 1996 are rather marginal;
4 Ya.V. Pavlenkoand H.W.Duerbeck:The spectrum of V4334 Sgr in August, 1998
Nevertheless,the SEDofV4334Sgr depends critically
on Teff (Pavlenko & Yakovina 2000). For the spectrum
of August 12, 1998, a best fit was obtained for Teff ∼
5250 K and EB−V = 1.3 (Fig. 4). The fit is based on the
Asplund et al. (1999) abundances for October 1996. The
corresponding values derived for our spectrum of April
29, 1997, are Teff = 5500 K and EB−V = 0.7 (Pavlenko
et al.2000).In contrastto the rapidcoolingof V4334Sgr
in 1996, Teff has apparently declined by only ∼ 250 K
1
in the course of the 15 months between early 1997 and
2
mid-1998.
4. Discussion
Our SEDs were computed in the plane-parallel approach.
This may be used for Teff = 5500 K (see discussions in
Asplund et al. 1998 and Pavlenko et al. 2000),but model
Fig.3. SEDs computed for model atmospheres 5200/0.0 atmospheres for Teff =5250K are more likely affected by
withabundancestakenfromAsplundetal.(1999),Kipper sphericity effects. Fits of model atmospheres with logg =
& Klochkova(1997)and the “carbon-rich”case.The first 0 and logg = 1 to the observed spectrum have a similar
two SEDs nearly coincide. quality.This,however,onlyindicatesthatthedependence
of the model on gravity is rather weak.
It is noteworthy that our theoretical fluxes are much
too high between 600 and 700 nm, which suggests that
an important molecular line opacity is missing. A similar
effect was noted by Asplund et al. (1997) in R CrB for λ
500 nm. It is doubtful that the assumption of geometry
and homogeneity is able to produce a shortage of flux at
suchaveryspecificwavelengthinterval.InV4334Sgr,the
effectisprobablycausedbytheabsorptionofoneormore
molecules which are not yet identified.
The observed and computed spectra for August, 1998
aremorediscrepantinthe blue region,while thefitinthe
red looksacceptable.A similar phenomenon was found in
the investigation of the April 1997 spectrum. By August
1998thedustenvelopehadbecomethick,whileitwasab-
sentormarginalinApril1997.Thusweconcludethatthe
discrepancies are likely not caused by dust, but by one
of the other effects outlined above. In the frame of our
simple model, which assumes that the circumstellar and
the interstellar dust have similar properties in the visible
Fig.4. Synthetical spectra for different effective temper-
atures and logg =0 are superimposed onthe spectrum of partof the spectrum,EB−V increasedfrom0.7 to 1.3,ac-
cordingto ourspectroscopicanalysis.Apparently,in1997
V4334 Sgr (thick line) observed in August, 1998.
there was no “active” optical circumstellar reddening in
the line of sight, which is proven by the similar effective
temperatures derived from spectroscopy and photometry
TheoverallshapeofthespectrumofV4334Sgrisgov- (Fig. 2 of Asplund et al. 1997), while at that time some
erned by the bands of the C2 Swan system and by the dust clouds may have already formed outside the line of
violet and red systems of CN (Pavlenko et al. 2000). In sight,inordertoaccountforthe IRexcessasobservedby
the near UV (λ < 400 nm), atomic absorption becomes Kimeswenger et al. (1997) and Kamath & Ashok (1999).
important (see Pavlenko & Yakovina 2000). The increase in EB−V between April 1997 and August
Computed and observed spectra were normalized to 1998, ∆EB−V = 0.6, is thus caused by circumstellar red-
equal flux at λ ∼ 740 nm. In general, our main conclu- dening of newly formed dust. If the optical properties of
sions show a rather weak dependence on the choice of λ this material are similar to those of interstellar material,
normalization (see Pavlenko et al. 2000). it should cause a drop in V by 0.6 × 3.1 ∼ 1.85 mag
Ya.V. Pavlenkoand H.W.Duerbeck:The spectrum of V4334 Sgr in August, 1998 5
Fig.5. Fits to SEDs with model 4000/0.0 of different Fig.6. Fits to SEDs of August 12, 1998 dereddened for
abundances. The expected very strong carbon bands can EB−V = 1.0 and 1.3 with model 5250/0.0of the “carbon-
be decreased in strength by reducing the C/O ratio, but rich”case.Therearesomeobviousproblemsofthefittings
the overallfit would be still poor. in the red and in the blue.
9000
(with an uncertainty of ±0.1) with respect to a “dust-
T
free” lightcurve. eff
In order to estimate the magnitude of V4334 Sgr in
8000
such a “dustfree” lightcurve, the trend in the 1996-97 V
light curve (as taken from Duerbeck et al. 2000) was lin-
early extrapolated to August, 1998, and an “expected”
7000
magnitudeofV =10.75wasderived;ifonlythe1997data
are used for an extrapolation, the result is V = 10.65. In
fact, the object is observed at V = 12.55. If we trust in
the validity of the linear extrapolation, the circumstellar 6000
A , which is necessary to put the object to the observed
V
brightness,is1.85±0.05,ingoodagreementwiththevalue
expected from the result of the spectroscopic study. 5000
Isthereanotherwaytoexplainthe observedspectrum
ofAugust, 1998,e.g.by keeping the reddening at its orig- 1996 1997 1998 1999
inalvalue ofEB−V =0.7,anddecreasingthe surfacetem- Year
perature? If Teff is assumed to be as low as 4000 K, a
generalagreementwith the overallSEDmaybe achieved,
but the fit to the molecular bands is extremely poor. The Fig.7. Temporal changes of Teff in V4334 Sgr. Open
circles: Asplund et al. (1997), open squares: Asplund et
fitcannotbe improved,eveniftheabundancesofC,N,O
al. (1999), open diamond: Kipper and Klochkova (1997),
arepermittedtovary(Fig.5).Hereweshowcomputations
filled circle: Pavlenko et al. (2000), filled square: this pa-
for differentC/O ratios(giveninlogarithmic scales).If O
per.
is more abundant than C, a completely different molecu-
lar chemistry, and therefore bands of oxygen compounds
would appear in V4334 Sgr. Fig. 5 illustrates that any
changesofcarbonabundancesdonotresultinabetter fit
of the observed spectrum. uncertainties in these estimates. As we see in Fig. 6, our
The estimate of Teff depends on the assumed EB−V. fitwithEB−V =1.0lookslessperfectbothinthe redand
Fig.6showssomefitsofcomputedspectratotheobserved intheblue bothfortheAsplundatal.(1999)abundances
spectrumofV4334Sgr,dereddenedforEB−V =1.0.There forMay1996andevenforthe“carbon-rich”case.Neither
we also plot the SED of V4334 Sgr, computed for the variations of logN(C) nor of EB−V permit to obtain a
“carbon-rich” case. This permits us to better assess the better fit to the SED observed in August, 1998.
6 Ya.V. Pavlenkoand H.W.Duerbeck:The spectrum of V4334 Sgr in August, 1998
Thus we conclude that Teff of V4334 Sgr was near Kamath, U.S., Ashok,N.M. 1999, MNRAS302, 512
5250 K in August, 1998, and that the overall modifica- Kerber, F., Palsa, R., K¨oppen, J., Bl¨ocker, T., Rosa, M.R.
tion of the SED is caused by circumstellar dust. 2000, ESO Messenger, No. 101, 27
The temporal changes of Teff during 1996-1998 are Kimeswenger, S., Gratl, H., Kerber, F., Fouqu´e, P., Kohle, S.
Steele, S. 1997, IAUCirc. 6608
showninFig.7.Following the assumptionofDuerbeck et
Kipper, T., Klochkova,V. 1997, A&A 324, L65
al.(1998),that changesof Teff were causedby the growth
Liller, W., Janson, M., Duerbeck, H.W., van Genderen, A.M.
of the pseudophotosphere surrounding a mass-losing ob-
1998, IAU Circ. 6825
ject that radiates at approximately constant luminosity,
Nakano, S., Sakurai, Y., et al. 1996, IAU Circ. 6322
we conclude that its expansion had noticeably slowed
Pavlenko,Y.V.,Yakovina,L.A.1994, Astrophys.Rep.38, 768
down or had even almost come to a halt during 1997 Pavlenko, Ya.V.1997, ApSS253, 43
– 1998, and its temperature had almost stabilized. This Pavlenko, Ya.V.1999, Astron. Reports43, 94
stagewasaccompaniedbydustformationprocessesinthe Pavlenko,Ya.V.,Yakovina,L.A.2000,Astron.Reports44,209.
envelopeabovethestellarphotosphere,whosestrengthin- Pavlenko, Ya.V., Yakovina,L.A., Duerbeck,H.W. 2000, A&A
creased with time. The object recoveredonly partly from 354, 229
theseevents,sothatbythesecondhalfof1999thephoto- Pavlenko, Ya.V.2000, ApSS,submitted.
Piskunov, N.E., Kupka, F., Ryabchikova, T.A., Weiss, V.V.,
sphere of V4334Sgr had become basically invisible in the
Jeffery,C.S. 1995, ApSS112, 525
optical region.
Pollacco, D. 1999, MNRAS 304, 127
5. Conclusions
A comparison of the observed spectra of V4334 Sgr in
April,1997andAugust,1998withcomputedSEDsshows
thatthetemperatureoftheradiatingatmospherehadonly
decreasedinTeff by 250K,while the EB−V hadincreased
by 0.6. This appears to be in perfect agreement with the
observeddepressioninthevisuallightcurve,ifastandard
interstellarextinctionlawAV =3.1×EB−Visassumedfor
newly formed circumstellar dust. Thus there is not only
photometric, but also spectroscopic evidence that (a) the
light curve depressions are caused by dust, and that (b)
the luminosity and mass loss changed in such a way that
the pseudophotosphere was kept at an almost constant
effective temperature in 1997 – 1998.
Acknowledgements. We cordially thank Dr. A. Piemonte for
taking the spectrum, Prof. T. Kipper and Dr. M. Asplund
for useful discussions, and the VALD database team for help-
ful service. Partial financial support of the investigations of
Ya. Pavlenko was provided by a Small Research Grant of the
American Astronomical Society.
References
Asplund,M., Gustafsson, B., Lambert, D.L., Rao, N.K. 1997,
A&A 321, L17
Asplund, M., Gustafsson, B., Kameswara Rao, N., Lambert,
D.L. 1998, A&A332, 651
Asplund, M., Lambert, D.L., Kipper, T., Pollacco, D.,
Shetrone, M.D. 1999, A&A343, 507
Asplund,M.,Gustafsson,B.,Lambert,D.L.,Rao,N.K.2000,
A&A 353, 287
Clayton, G.C., deMarco O. 1997, AJ 114, 2679
Duerbeck,H.W., Benetti, S. 1996, ApJ 468, L111
Duerbeck,H.W.,vanGenderen,A.,Jones,A.,Liller,W.1999,
Southern Stars 38, 80
Duerbeck, H.W., Liller, W., Sterken, C. et al. 2000, AJ 119,
2360
Hofsaess, D. 1979, Atomic data and nuclear data tables, 24,
285